A method for measuring orbital volume using low-dose CT with contiguous 3mm transaxial sections is described. The accuracy of the method is 1.6%, ...
Orbital volume measured by a low-dose CT scanning technique M. McGurk, R.W. Whitehouse*, P.M. Taylor* and B. Swinson" Department of Oral and Maxillofacial Surgery, Queen's Medical Centre, University Hospital, Nottingham. "Department of Diagnostic Radiology, University of Manchester and 'Department of Oral and Maxillofacial Surgery, Turner Dental School, Manchester, UK
Received 9 March 1991 and in final form 20 October 1991 A method for measuring orbital volume using low-dose CT with contiguous 3 mm transaxial sections is described. The accuracy of the method is 1.6%, as demonstrated by comparing CT volume measurements with those derived directly from alginate impressions and on repeat scanning the precision of the measurement was judged as 1.3%. Within the same individual, the right and left orbital volumes were observed to be within 0.6 cm' (s.d. ±0.33 ern") of each other. This study demonstrates that low-dose CT scanning is a practical method of determining orbital volume and could be used to advantage in the management of traumatic enophthalmos and blow-out fractures of the orbit. Keywords: Tomography, X-ray computed; orbit; fractures; maxillofacial injuries, enophthalmos Dentomaxillofac. Radial., 1992, Vol. 21, 70-72, May
Computed tomography (CT) is now an established method of evaluating craniofacial injuries following trauma and is particuarly useful in orbital injuries, as it can be used to distinguish both osseous and soft-tissue damage I. Forbes and colleagues" have refined its use to allow the volume of the orbit as a whole or the softtissue compartments alone (that is, intraconal and extraconal fat) to be calculated. Only the total orbital volume was found to be important in post-traumatic enophthalmos. The disadvantage of conventional volume measurement is that thin section, high-dose CT scans are employed, with an attendant increase in radiation exposure to the patient, on the assumption that high definition images are required for accuracy. In addition, off-line computing has been needed to calculate the volumes from the digital data. The facility to measure volume is relevant to the management of traumatic enophthalmos, since its pathogenesis is principally related to an increase in orbital volume. The conventional method of treating this disorder is to pack the floor of the orbit with bone or alloplastic material in an attempt to restore the orbit to its original volume. However, the quantity of graft required to meet this objective is judged empirically and, as a consequence, the results have been unpredictable. The advantage of pre-operative radiological volume quantification is that it permits the surgeon to tailor the size of the graft to restore the orbit accurately to its original volume, using the uninjured orbit as the standard. The present study reports a simple method for determining orbital volume from a low-dose CT scan without the requirement for additional computing facilities. The method has been validated by comparing CT-generated volumes with those obtained by direct orbital measurement. 70
Dentomaxillofac. Radiol., 1992, Vol. 21, May
Materials and methods Nineteen dried skulls of various sizes and with intact bony orbits were provided by the Department of Anatomy, University of Manchester. Seven skulls were scanned using a low-dose technique (120kVp, 40mA, 2s scan time, 3mm contiguous transaxial sections) on a GE 9800 General Purpose CT Scanner (I.G.E. Medical Systems Ltd, Slough, UK). The volume of the orbit was calculated by summating the pixel counts in successive cross-sections through the orbit. Three millimetre contiguous CT slices provided a series of planes for orbital area measurement and the sum of the areas was multiplied by the slice thickness to provide the volume. The radiological boundaries of the orbit were defined anteriorly by a line connecting the anterior surface of the zygomaticofrontal process to the nasomaxillary suture and posteriorly by the optic foramen (Figure 1). All the measurements were performed on the CT console at a fixed window width and level. The CT scanning and orbital volume estimations were performed independently of the direct orbital measurements. The same scanning procedure and volume measurement was repeated on five skulls after an interval of 12 months by a second observer to estimate the precision of measurement and to assess interobserver variation. This scanning protocol has been previously shown by thermoluminescent dosimetry to have a skin entry dose of 11mGy (ref. 3). The orbital volumes of 19 dried skulls (38 orbits), including the scanned specimens, were measured directly by filling each orbit with alginate impression material to a plane joining the superior and inferior margins of the orbit and the nasomaxillary suture. This plane does not necessarily coincide with the anterior boundary of the orbit described radiologically.
Orbital volume received by CT: M. McGurk et al. r----------------------,
25
,., e u GJ
E
:l
20
(3
>
....
•
o
15 30
25
Alginate volume Icm 3) Figure 2 Comparison of the measurement of orbital volume by computed tomography and alginate impression in seven skulls. The correlation between the two techniques was high (r = 0.96) with a regression line slope of I. -. R ight orbit; D. left orbit 33 r - - - - - - - - - - - - - - - - - - - - ,
Figure 1 A representative low-dose cr section through the orbits showing the line marking the anterior limit of the orbit. The area of interest mea sured by the scanner is designated 1. The volume of the orb it is der ived by summating the volume within the traced boundary of the orbit in each consecutive section
,.,
e
u
GJ
e:l (3
>
Alginate is a hydrocolloid with elastic properties that regains its shape after deformation and therefore preserves an accurate record of the orbit after removal. The orbital foramina were sealed with wax and moistened gauze strips (1 cm x 3 cm) were placed in the medial and lateral walls of the orbit to aid in the withdrawal of the alginate impression. The impression material (Kromogel, Wright Health Group Ltd , Dundee, Scotland) was of a set composition (16.1 g alginate to 39 ml of water) and was mixed to a uniform consistency on a vibrating table as tested by weighing set volumes of different mixes. The mean density of the mixes was 1.096 g crrr' (s.d. 0.003). The volume of the impression was calculated by Archimedes principle as: Volume
= Mass/Density
Results The mean volume determined by the direct impression of 38 orbits was 28.67cm 3 (range 25.56-31.78cm 3 ) . The volumes derived from the CT scans in 14 orbits are compared with those derived directly in Figure 2. The correlation between the two techniques was high (r = 0.96) with a regression line slope of 1. The accuracy of this measurement method was 1.6% (calculated as the standard error of the estimate from linear regression between CT measured and actual volume, expressed as a percentage of the mean volume) . There was a consistent volume discrepancy between the two methods with CT underestimating the alginate volume measurements by an average of 8.7 em:' for each orbit (range 7.86-9.43cm 3 ) . Long-term reproducibility for repeat scanning by the same observer was 1.3% (coefficient of variation) and repeat analysis of the same scans by two observers showed an interobserver error of 1.6% (coefficient of variation) . For all 19 skulls measured by alginate impression , the mean difference in volume between the
• 1Il.
•
31
.
•
:0 0
.......
•
29
~
27
GJ
--'
1Il
•
•
· -.
•• •
25 25
27
29
Right orbital volume Icm
31
33
3)
Figure 3 Comparison of right and left orbital volumes measured by alginate impression in 19 skulls (38 orbits)
orbits in the same individual was 0.6 cm' (s.d. 0.33 cm') and was not lateralized (Figure 3).
Discussion The orbit is an excellent subject for computerized volume study, as marked differences in contrast exist between the intra- and peri -orbital structures (fat, muscle, fluid, air and bone) that minimize the overlap of density ranges for determining boundaries. Consequently, it has proved possible, using thin section, highdefinition CT, not only to measure orbital volume but also that of the muscles and fat within the orbit", However this technique has several disadvantages, the large number of images produced, the increased scanning and computing time needed for analysis and the dose of radiation received by the patient. The latter is important because the lens of the eye is particularly sensitive to radiation damage, with detectable lens opacities occurring after single exposures of as little as 500mSv4 . Forbes et al. 2 and BallS have described protocols for scanning the orbit where the dose of radiation received by the eye was between 35 and 60mSv. In many patients, however, the scans were taken in multiple planes resulting in a total lens dose of up to 105 mSv. This is close to the recommended annual dose-equivalent limit for members of the public of 150 mSv (ref. 4) . The present study indicates that Dentomaxillofac. Radiol., 1992, Vol. 21, May
71
Orbital volume received by CT: M. McGurk et al. low-dose CT of the orbit reduces the radiation dose to approximately 11 mSv, yet maintains image quality sufficient to permit accurate orbital volume measurement. The technique does not significantly affect the detection of bone edges" nor the demonstration of softtissue abnormality within the orbits". The low X-ray output also reduces the heat loading of the CT scanner and the examination can therefore be performed in 'dynamic mode', allowing all the images to be acquired within 2 min, with the advantage of increasing patient throughput and limiting the potential for patient movement". The reduction of movement during scanning is important as it could cause marked error in orbital volume estimation. Previous studies on our scanner have demonstrated the need for tube cooling pauses after every 10 sections with exposures of 200 mAs or higher in dynamic mode, whereas the lowdose technique needs no such pauses", The short scanning time also has the advantage that patients can be easily fitted in to busy scan schedules and consequently surgery is not delayed. Whilst computerized edge detection methods using thin section CT images may be considered the gold standard for orbital volume estimationv ':", the results of the current study show equivalent accuracy for a manual technique requiring no additional computing facilities or software. The reproducibility of the technique is excellent and the measurement takes less than 10 min to complete. We have used this low-dose method successfully for a number of years to assess patients with orbital trauma and have found the quality of the images adequate for identifying both bony and soft-tissue abnormalities. A formal review of over 60 cases is currently in progress. The present study validates the use of the low-dose transaxial CT technique for volume measurement in this group of patients. It is difficult with radiological images to identify an anterior limit to the orbit that coincides exactly with that judged clinically, and therefore absolute measurements of orbital volume are not readily obtained from CT data. In the present study, orbital volume generated with CT was approximately 8.6 ern" less than those derived by direct measurement primarily for this reason. An additional source of the underestimation is partial volume averaging of the orbital content with the bone edge, the inner aspect of which was always used to define the orbit in this technique. One skull was rescanned using 1.5 mm sections and the volume measurements repeated to assess this effect. The volume was found be to 1 crrr' greater than with the 3 mm section technique but the difference between right and left orbits was unchanged. As a similar (though smaller) partial volume averaging error will also be present with 1.5 mm sections, this source of discrepancy may account for up to one-quarter of the underestimation. The fact that measurements derived from 3 mm low-dose CT scan data record volumes accurately is indicated by the high coefficient of correlation of r = 0.96 and the regression line slope of 1. In fact, 5 mm sections have been used without any
72
Dentomaxillofac. Radiol., 1992, Vol. 21, May
obvious loss of accuracy", The mean CT volume of the orbits in this study compares favourably with measurements reported by others":", The present data indicate that the volumes of the orbits in anyone skull are normally within 0.6 crrr' of each other, consistent with previous studies/v':". A relatively modest displacement of either the medial wall or floor of the orbit (3 mm) can be associated with changes in orbital volume of 1.2-2.6 em" (ref. 9) that are capable of inducing enophthalmos". The fact that the volume changes responsible for enophthalmos generally exceed, by a factor of two, the mean volume difference between right and left orbits permits the normal orbit to be used as a standard by which to judge the traumatized socket. The data suggest that a volume difference of greater than 1.2cm3 (mean +2s.d.) between orbits can be considered abnormal. The present study was designed to provide a practical method of investigating traumatic enophthalmos, the pathogenesis of which relates to a relative increase in volume in the traumatized orbit 10. Currently, it is not routine surgical practice to establish the volume change responsible for the enophthalmos prior to surgery and at operation the quantity of graft material packed into the orbit is judged empirically. It is hoped that by accurately restoring the volume of the traumatized orbit with measured volumes of graft material, the results of surgery will be improved.
References I. Gilbard SM. Mafee MF. Lagouros PA. Langer BG. Orbital blow out fractures. Ophthalmology 1985; 92: 1523-8. 2. Forbes G. Gehring D. Gorman CA. Brennan MD. Jackson IT. Volume measurement of nasal orbital structures by computer tomographic analysis. Am J Neuroradiol1985; 145: 149-54. 3. Cooper We. A method for volume determination of the orbit and its contents by high resolution axial tomography and quantitative digital image analysis. Trans Am Ophthalmol Soc 1985; 83: 547-609. 4. ICRP Publication 41. Nonstochastic effects of ionising radiation. Ann lCRP 1984; 14: 17-33. 5. Ball JB. Direct oblique sagittal CT of orbital wall fractures. AJR 1987; 148: 601-8. 6. Gholkar A. Gillespie JE. Hart CWo Mott D. Isherwood I. Dynamic low-dose three-dimensional computed tomography: a prelimary study. Br J Radial 1988; 61: 1095-9. 7. Whitehouse RW. Leatherbarrow B. A craniocerebral erosion (growing skull fracture) causing anisometropia. Br J Radio11990; 63: 728-30. 8. Bite U. Jackson IT. Forbes GS. Gehring DG. Orbital volume measurements in enophthalmos using three-dimensional CT imaging. J Plast Reconstr Surg 1985; 75: 502-507. 9. Parsons S. Mathog RH. Orbital wall and volume relationships. Arch Otolaryngol Head Neck Surg 1988; 114: 743-7. 10. Hammerschlag SB. Hughes S. O'Reilly GV. Weber AL. Another look at blow-out fractures of the orbit. AJR 1982; 139: 133-6. Address: Dr M. McGurk. Department of Oral and Maxillofacial Surgery. Queen's Medical Centre. University Hospital. Nottingham NG7 2UH, UK.